Onnx runtime and ML2

General
1

I'm trying to use MS's onnx runtime packages on a project for the ML2. I'm using ORT becuase barracuda doesn't work for my model type. The code works in Unity, but the session (create by InferenceSession(), throws a null exception when I try to use Run(). This only happens on device and not in the Unity editor. Any ideas?

Specifically at this line:
results = session.Run(inputs).First().AsEnumerable();

I get this exception:
Object Reference Not Set To An Instance Of An Object
2023-07-17 21:37:15.651;10106;16270;16290;Info;Unity;at System.Linq.Enumerable.First[TSource] (System.Collections.Generic.IEnumerable`1[T] source) [0x00000] in <00000000000000000000000000000000>:0;

Had a thought. It may be the the onnx file isnt being loaded propely. What directory does the file need to be in to be loaded?

2

Unfortunately, we are not familiar with this solution and recommend asking on the MS's Onnx Runtime Package's github.

You can also test if the issue is specific to Magic Leap 2 or Android, by testing the application on the Android Studio Simulator. Simply create a Unity 2D application that targets x86_64 and deploy it to the virtual device. Make sure the device is configured for ( Android 10, x86_64 , No Google Play services). If your application runs as expected, you can then deploy the 2D app to your headset and see if the issue persists.

3

OK. Is there a recommended way to run an SVM model on the ML2?

4

I'm not sure I understand, SVM model?